We designed a systematic study of patients with Cushing's disease to compare the results of acute experiments with cyproheptadine, sodium valproate and bromocriptine with the results of chronic treatment with sodium valproate. In 13 patients the plasma cortisol response to single doses of 2.5 mg bromocriptine, 6 mg cyproheptadine and 300 mg sodium valproate was assessed over 4-8 h. All patients were then treated with sodium valproate 200 mg three times a day for a period of 3 months. Subjects were classified as responders when there was a reduction in plasma cortisol by at least 50% below the basal level in two consecutive plasma samples during one of these tests, and/or clinical remission and normalization of cortisol production rate during sodium valproate treatment. Four patients responded significantly to at least one of the agents (group I) and 9 did not (group II). In group I three patients were bromocriptine responsive and one of these responded also to cyproheptadine; in this patient both sodium valproate and cyproheptadine were able to induce a remission during chronic treatment. In the one patient of the responder group who was bromocriptine insensitive, sodium valproate treatment also induced a remission. In two patients of the responder group sodium valproate treatment was ineffective. However, in both patients chronic treatment with bromocriptine induced a marked clinical improvement associated with a decrease of cortisol production rate. There was no difference between the groups in sex, age, clinical presentation, duration or severity of hypercortisolism. In one patient in group I a macroadenoma with suprasellar extension was present, while a microadenoma was detected in four patients in group II.(ABSTRACT TRUNCATED AT 250 WORDS)